http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
통계적 형상밀도 모델을 이용한 족부 골격의 엑스선 영상 정합
오정석(Jeongseok Oh),구승범(Seungbum Koo) 대한기계학회 2021 대한기계학회 춘추학술대회 Vol.2021 No.11
Accurate kinematics measurement of foot bones is necessary for a profound understanding of the pathomechanics of bone injuries such as plantar fasciitis and flatfoot. Bi-planar fluoroscopy has been used for kinematics measurements. Previously, edge-based 2d-to-3d registration methods have been used to find the three-dimensional pose of the bone, which did not utilize the bone density information in radiographs for the registration. In this study, we developed a method to obtain the three-dimensional pose, shape, and intensity of the foot bones without subject specific bone models by using a statistical shape and intensity model (SSIM). A digitally reconstructed radiograph of the SSIM was used for registering both pose, shape, and intensity. The accuracies in predicting three-dimensional pose were evaluated for the tibia, talus, and calcaneus. The registration errors along the x-, y-, and z-axes were 0.87, 0.74, and 1.10 mm, respectively, and the median Hausdorff distance of the registered model was 2.07 mm.
다중 두 방향 엑스선 영상을 이용한 무릎 골격의 형상밀도 예측
오정석(Jeongseok Oh),윤승우(Seungwoo Yoon),구승범(Seungbum Koo) 대한기계학회 2022 대한기계학회 춘추학술대회 Vol.2022 No.11
Bi-planar fluoroscopic (BPF) image processing has been widely used for measuring the skeletal kinematics of several joints. The procedure of estimating the 3-dimensional (3D) kinematics from 2-dimensional (2D) BPF images is called 2D-3D registration. The reconstruction of the subject-specific bone geometry is essential in 2D-3D registration. The studies to find the bone geometry without using the CT scans which automatically estimate the shape as well as the pose were investigated. In this study, we propose the reconstruction of the distal femur by using a series of BPF images and a statistical shape and intensity model and checked the effect of the number of frames to be used for the optimization of the accuracy of the reconstruction. Three CT scans of cadaveric knee joints were prepared to quantify the accuracy of reconstruction. The cadaveric knee joints were adjusted to simulate knee flexion while the fluoroscopic imaging system was taking the X-ray images. For every time stamp, both bi-planar images were synchronized and calibrated. The Hausdorff distance between the reconstructed bone and ground-truth bone was used for representing the accuracy of the reconstruction. While the number of frames to be used for the reconstruction was changed, the Hausdorff distance was compared. The accuracy of reconstruction decreased when using a larger number of frames and reached 4.23±0.53 mm on average when using 6 frames of BPF image pairs.
다중 두 방향 엑스선 영상을 이용한 무릎 골격의 형상밀도 예측
오정석(Jeongseok Oh),윤승우(Seungwoo Yoon),구승범(Seungbum Koo) 대한기계학회 2022 대한기계학회 춘추학술대회 Vol.2022 No.11
Bi-planar fluoroscopic (BPF) image processing has been widely used for measuring the skeletal kinematics of several joints. The procedure of estimating the 3-dimensional (3D) kinematics from 2-dimensional (2D) BPF images is called 2D-3D registration. The reconstruction of the subject-specific bone geometry is essential in 2D-3D registration. The studies to find the bone geometry without using the CT scans which automatically estimate the shape as well as the pose were investigated. In this study, we propose the reconstruction of the distal femur by using a series of BPF images and a statistical shape and intensity model and checked the effect of the number of frames to be used for the optimization of the accuracy of the reconstruction. Three CT scans of cadaveric knee joints were prepared to quantify the accuracy of reconstruction. The cadaveric knee joints were adjusted to simulate knee flexion while the fluoroscopic imaging system was taking the X-ray images. For every time stamp, both bi-planar images were synchronized and calibrated. The Hausdorff distance between the reconstructed bone and ground-truth bone was used for representing the accuracy of the reconstruction. While the number of frames to be used for the reconstruction was changed, the Hausdorff distance was compared. The accuracy of reconstruction decreased when using a larger number of frames and reached 4.23±0.53 mm on average when using 6 frames of BPF image pairs.